From b833a8d25556e03b0c56d306798993472a32f0f2 Mon Sep 17 00:00:00 2001
From: Hypolite Petovan <mrpetovan@gmail.com>
Date: Sun, 22 Oct 2017 23:48:00 -0400
Subject: [PATCH] Fix cron_maintain

- Update profile before fetching scrape url
- Honor $parms['hide'] before validating dfrn site
- Fetch maintenance items oldest first
- Add backlog size in log
- Add pid to logger for easier threaded cron debug
---
 boot.php                  |  2 +-
 include/cron_maintain.php | 21 ++++++++++++++++-----
 include/submit.php        | 24 +++++++++++++++++-------
 3 files changed, 34 insertions(+), 13 deletions(-)

diff --git a/boot.php b/boot.php
index 0f4f36a4..9944e881 100644
--- a/boot.php
+++ b/boot.php
@@ -79,7 +79,7 @@ if (!function_exists('logger')) {
 		}
 		require_once('include/datetime.php');
 
-		@file_put_contents($logfile, datetime_convert() . ':' . ' ' . $msg . "\n", FILE_APPEND);
+		@file_put_contents($logfile, datetime_convert() . ' [#' . getmypid() . '] ' . $msg . "\n", FILE_APPEND);
 		return;
 	}
 }
diff --git a/include/cron_maintain.php b/include/cron_maintain.php
index 960fe901..0e1051d4 100644
--- a/include/cron_maintain.php
+++ b/include/cron_maintain.php
@@ -21,12 +21,23 @@ require_once '.htconfig.php';
 require_once 'dba.php';
 $db = new dba($db_host, $db_user, $db_pass, $db_data, $install);
 
-//Get our set of items. Youngest items first, after the threshold.
-//This may be counter-intuitive, but is to prevent items that fail to update from blocking the rest.
+
+//Get the maintenance backlog size.
+$res = q("SELECT count(*) as `count`
+FROM `profile`
+WHERE `updated` < '%s'",
+	dbesc(date('Y-m-d H:i:s', time() - $a->config['maintenance']['min_scrape_delay']))
+);
+$maintenance_backlog = 'unknown';
+if (count($res)) {
+	$maintenance_backlog = $res[0]['count'] . ' entries left';
+}
+
+//Get our set of items. Oldest items first, after the threshold.
 $res = q("SELECT `id`, `homepage`, `censored`
 FROM `profile`
 WHERE `updated` < '%s'
-ORDER BY `updated` DESC
+ORDER BY `updated` ASC
 LIMIT %u",
 	dbesc(date('Y-m-d H:i:s', time() - $a->config['maintenance']['min_scrape_delay'])),
 	intval($a->config['maintenance']['max_scrapes'])
@@ -56,9 +67,9 @@ $threads = array();
 
 //Debug...
 if ($verbose) {
-	echo "Creating $threadc maintainer threads for $items profiles." . PHP_EOL;
+	echo "Creating $threadc maintainer threads for $items profiles, $maintenance_backlog"  . PHP_EOL;
 }
-logger("Creating $threadc maintainer threads for $items profiles.");
+logger("Creating $threadc maintainer threads for $items profiles. $maintenance_backlog");
 
 for ($i = 0; $i < $threadc; $i++) {
 
diff --git a/include/submit.php b/include/submit.php
index 255657b4..90876c2d 100644
--- a/include/submit.php
+++ b/include/submit.php
@@ -29,6 +29,14 @@ function run_submit($url) {
 	if(count($r)) { 
 		$profile_exists = true;
 		$profile_id = $r[0]['id'];
+
+		$r = q("UPDATE `profile` SET
+			`updated` = '%s'
+			WHERE `id` = %d LIMIT 1",
+
+			dbesc(datetime_convert()),
+			intval($profile_id)
+		);
 	}
 	
 	//Remove duplicates.
@@ -86,19 +94,21 @@ function run_submit($url) {
 		nuke_record($url);
 		return true; //This is a good update.
 	}
-	
-	//This is most likely a problem with the site configuration. Ignore.
-	elseif(validate_dfrn($parms)) {
-		return false;
-	}
-	
+
 	if((x($parms,'hide')) || (! (x($parms,'fn')) && (x($parms,'photo')))) {
 		if($profile_exists) {
+			logger('Profile inferred to be opted out of the directory.');
 			nuke_record($url);
 		}
 		return true; //This is a good update.
 	}
-	
+
+	//This is most likely a problem with the site configuration. Ignore.
+	if(validate_dfrn($parms)) {
+		logger('Site is unavailable');
+		return false;
+	}
+
 	$photo = $parms['photo'];
 
 	dbesc_array($parms);